Use the corpus

Take the texts. They were never ours to keep.

Everything Wellspring reads from is published as plain JSON at a stable address, free to download, mirror, query and build on. You need no key, no account and no permission from us — only to honour what each text's own licence asks. This page is the shape of it.

What is there

Eighty packages: 28 Bibles and ancient versions, 33 works, 5 commentaries, 4 dictionaries, 4 lexicons, and the indexes computed over them — cross-references, topics, places, concepts and reading plans. 748,493 verses in all. Forty-nine packages are public domain. Of the thirty-one that are not, twenty-three are share-alike and six are CC BY; two state their own terms, and the register gives them in full.

The library page lists every work with its edition, language and rights, generated from the register so it can never claim a text the shelf does not hold. Each package states its own rights; there is no single licence over the whole, and anyone redistributing should read them one by one.

Where it is

The register first — it names every package and what is in it:

https://tudoxukno.github.io/wellspring-corpus/content/manifest.json

Then each package, by the id the register gives it:

…/content/<package>/index.json          the books, chapters and counts
…/content/<package>/source.json         title, language, rights, citation, method
…/content/<package>/books/<OSIS>.json   the text, chapter by chapter and verse by verse
…/content/<package>/text.json           the same text flat, for searching

So the King James book of Daniel is …/content/kjv/books/Dan.json, and the Ge'ez is …/content/geez/books/Dan.json. Books are named by their OSIS abbreviation. Nothing is paginated and nothing is behind a query parameter.

What the computed layers hold

These are ours, derived from the texts, and they are the part least likely to exist anywhere else free:

What we ask

Nothing beyond what the licences themselves require: attribution where a licence asks for it, and share-alike where a licence asks for that. We ask no attribution of our own and set no extra condition. If a package's rights note is unclear or looks wrong to you, tell us and we will fix the register — that is the only thing here we would call a favour.

One restriction worth knowing before you redistribute. The Sahidic Coptic New Testament (sahnt, Sahidica, © 2000–2006 J. Warren Wells) is free of charge in free electronic editions with its full title and copyright credited, but print requires written permission. It is the one package on the shelf where taking the whole corpus and putting it on paper would be a breach. The reading plans are mixed and state their terms one by one.

Two requests that are courtesy rather than licence. Please mirror rather than hot-link if you are serving many readers, so a small project's bandwidth is not holding up yours. And please do not present a scanned text as clean: several works came off OCR and carry their own account of what the scanner got wrong, in docs/repairs/ in the repository.

The register is public too

Every hold, every rights decision and every ingest script is at github.com/tudoxukno/wellspring-corpus. If you want to know why a text is on the shelf, or why one you expected is not, the answer is written down there rather than decided privately.

If you are building something

You are welcome to, and you do not need to tell us. If you would like to, or if you want a package shaped differently — a flat CSV, a different chunking, a language we hold but you cannot parse — write to support@wellspringstudy.org. We would rather the texts were used than admired.